Continuous Glucose Monitors and Activity Trackers to Inform Insulin Dosing in Type 1 Diabetes: The University of Virginia Contribution

A strategy based on integer programming for optimal dosing and timing of preventive hypoglycemic treatments in type 1 diabetes management

BACKGROUND AND OBJECTIVES

One of the major problems related to type 1 diabetes (T1D) management is hypoglycemia, a condition characterized by low blood glucose levels and responsible for reduced quality of life and increased mortality. Fast-acting carbohydrates, also known as hypoglycemic treatments (HT), can counteract this event. In the literature, dosage and timing of HT are usually based on heuristic rules. In the present work, we propose an algorithm for mitigating hypoglycemia by suggesting preventive HT consumption, with dosages and timing determined by solving an optimization problem.

METHODS

By leveraging integer programming and linear inequality constraints, the algorithm can bind the amount of suggested carbohydrates to standardized quantities (i.e., those available in "off-the-shelf" HT) and the minimal distance between consecutive suggestions (to reduce the nuisance for patients).

RESULTS

The proposed method was tested in silico and compared with competitor algorithms using the UVa/Padova T1D simulator. At the cost of a slight increase of HT consumed per day, the proposed algorithm produces the lowest median and interquartile range of the time spent in hypoglycemia, with a statistically significant improvement over most competitor algorithms. Also, the average number of hypoglycemic events per day is reduced to 0 in median.

CONCLUSIONS

Thanks to its positive performances and reduced computational burden, the proposed algorithm could be a candidate tool for integration in a DSS aimed at improving T1D management.

Offline Deep Reinforcement Learning and Off-Policy Evaluation for Personalized Basal Insulin Control in Type 1 Diabetes

Recent advancements in hybrid closed-loop systems, also known as the artificial pancreas (AP), have been shown to optimize glucose control and reduce the self-management burdens for people living with type 1 diabetes (T1D). AP systems can adjust the basal infusion rates of insulin pumps, facilitated by real-time communication with continuous glucose monitoring. Deep reinforcement learning (DRL) has introduced new paradigms of basal insulin control algorithms. However, all the existing DRL-based AP controllers require extensive random online interactions between the agent and environment. While this can be validated in T1D simulators, it becomes impractical in real-world clinical settings. To this end, we propose an offline DRL framework that can develop and validate models for basal insulin control entirely offline. It comprises a DRL model based on the twin delayed deep deterministic policy gradient and behavior cloning, as well as off-policy evaluation (OPE) using fitted Q evaluation. We evaluated the proposed framework on an in silico dataset generated by the UVA/Padova T1D simulator, and the OhioT1DM dataset, a real clinical dataset. The performance on the in silico dataset shows that the offline DRL algorithm significantly increased time in range while reducing time below range and time above range for both adult and adolescent groups. Then, we used the OPE to estimate model performance on the clinical dataset, where a notable increase in policy values was observed for each subject. The results demonstrate that the proposed framework is a viable and safe method for improving personalized basal insulin control in T1D.

A New Index of Insulin Sensitivity from Glucose Sensor and Insulin Pump Data: In Silico and In Vivo Validation in Youths with Type 1 Diabetes

Background: Estimation of insulin sensitivity (SI) and its daily variation are key for optimizing insulin therapy in patients with type 1 diabetes (T1D). We recently developed a method for SI estimation from continuous glucose monitoring (CGM) and continuous subcutaneous insulin infusion (CSII) data in adults with T1D (SISP) and validated it under restrained experimental conditions. Herein, we validate in vivo a new version of SISP performing well in daily life unrestrained conditions. Methods: The new SISP was tested in both simulated and real data. The simulated dataset consists of 100 virtual adults of the UVa/Padova T1D Simulator monitored during an open-loop experiment, whereas the real dataset consists of 10 youths with T1D monitored during a hybrid closed-loop meal study. In both datasets, participants underwent two consecutive meals (breakfast and lunch, at 7 and 11 am) with the same carbohydrate content (70 g). Plasma glucose and insulin were measured during each meal to estimate the oral glucose minimal model SI (SIMM). CGM and CSII data were used for SISP calculation, which was then validated against the gold standard SIMM. Results: SISP was estimated with good precision (median coefficient of variation <20%) in 100% of the real and 91% of the simulated meals. SISP and SIMM were highly correlated, both in the simulated and real datasets (R = 0.82 and R = 0.83, P < 0.001), and exhibited a similar intraday pattern. Conclusions: SISP is suitable for estimating SI in both closed- and open-loop settings, provided that the subject wears a CGM sensor and a subcutaneous insulin pump.

Continuous Glucose Monitoring and Physical Activity

Continuous glucose monitoring (CGM) use has several potential positive effects on diabetes management. These benefits are, e.g., increased time in range (TIR), optimized therapy, and developed documentation. Physical activity is a recommended intervention tool in diabetes management, especially for people with type 2 diabetes (T2D). The benefits of physical activity for people with diabetes can be seen as an improvement of glycemic control, glycemic variability, and the reduction of insulin resistance. In relation to the physical activity of people with T2D, the benefits of CGM use can even be increased, and CGM can be a helpful tool to prevent adverse events due to physical activity of people with diabetes, such as hypoglycemic events and nocturnal hypoglycemia after sports. This narrative review aims to provide solid recommendations for the use of CGM in everyday life physical activities based on the noted benefits and to give a general overview of the guidelines on physical activity and CGM use for people with diabetes.

A Brief Review on the Evolution of Technology in Exercise and Sport in Type 1 Diabetes: Past, Present, and Future

One hundred years ago, insulin was first used to successfully lower blood glucose levels in young people living with what was then called juvenile diabetes. While insulin was not a cure for diabetes, it allowed individuals to resume a near normal life and have some freedom to eat more liberally and gain the strength they needed to live a more active lifestyle. Since then, a number of therapeutic and technical advances have arisen to further improve the health and wellbeing of individuals living with type 1 diabetes, allowing many to participate in sport at the local, regional, national or international level of competition. This review and commentary highlights some of the key advances in diabetes management in sport over the last 100 years since the discovery of insulin.

Simulation-Based Evaluation of Treatment Adjustment to Exercise in Type 1 Diabetes

Regular exercise is beneficial and recommended for people with type 1 diabetes, but increased glucose demand and changes in insulin sensitivity require treatment adjustments to prevent exercise-induced hypoglycemia. Several different adjustment strategies based on insulin bolus reductions and additional carbohydrate intake have been proposed, but large inter- and intraindividual variability and studies using different exercise duration, intensity, and timing impede a direct comparison of their effects. In this study, we use a mathematical model of the glucoregulatory system and implement published guidelines and strategies in-silico to provide a direct comparison on a single 'typical' person on a standard day with three meals. We augment this day by a broad range of exercise scenarios combining different intensity and duration of the exercise session, and different timing with respect to adjacent meals. We compare the resulting blood glucose trajectories and use summary measures to evaluate the time-in-range and risk scores for hypo- and hyperglycemic events for each simulation scenario, and to determine factors that impede prevention of hypoglycemia events. Our simulations suggest that the considered strategies and guidelines successfully minimize the risk for acute hypoglycemia. At the same time, all adjustments substantially increase the risk of late-onset hypoglycemia compared to no adjustment in many cases. We also find that timing between exercise and meals and additional carbohydrate intake during exercise can lead to non-intuitive behavior due to superposition of meal- and exercise-related glucose dynamics. Increased insulin sensitivity appears as a major driver of non-acute hypoglycemic events. Overall, our results indicate that further treatment adjustment might be required both immediately following exercise and up to several hours later, but that the intricate interplay between different dynamics makes it difficult to provide generic recommendations. However, our simulation scenarios extend substantially beyond the original scope of each model component and proper model validation is warranted before applying our in-silico results in a clinical setting.

Automatically accounting for physical activity in insulin dosing for type 1 diabetes

BACKGROUND AND OBJECTIVE

Type 1 diabetes is a disease characterized by lifelong insulin administration to compensate for the autoimmune destruction of insulin-producing pancreatic beta-cells. Optimal insulin dosing presents a challenge for individuals with type 1 diabetes, as the amount of insulin needed for optimal blood glucose control depends on each subject's varying needs. In this context, physical activity represents one of the main factors altering insulin requirements and complicating treatment decisions. This work aims to develop and test in simulation a data-driven method to automatically incorporate physical activity into daily treatment decisions to optimize mealtime glycemic control in individuals with type 1 diabetes.

METHODS

We leveraged glucose, insulin, meal and physical activity data collected from twenty-three individuals to develop a method that (i) tracks and quantifies the accumulated glycemic impact from daily physical activity in real-time, (ii) extracts an individualized routine physical activity profile, and (iii) adjusts insulin doses according to the prolonged changes in insulin needs due to deviations in daily physical activity in a personalized manner. We used the data replay simulation framework developed at the University of Virginia to "re-simulate" the clinical data and estimate the performances of the new decision support system for physical activity informed insulin dosing against standard insulin dosing. The paired t-test is used to compare the performances of dosing methods with p < 0.05 as the significance threshold.

RESULTS

Simulation results show that, compared with standard dosing, the proposed physical-activity informed insulin dosing could result in significantly less time spent in hypoglycemia (15.3± 8% vs. 11.1± 4%, p = 0.007) and higher time spent in the target glycemic range (66.1± 11.7% vs. 69.6± 12.2%, p < 0.01) and no significant difference in the time spent above the target range(26.6± 1.4 vs. 27.4± 0.1, p = 0.5).

CONCLUSIONS

Integrating daily physical activity, as measured by the step count, into insulin dose calculations has the potential to improve blood glucose control in daily life with type 1 diabetes.

Continuous Glucose Monitoring in Adults with Type 1 Diabetes: Real-World Data from the German/Austrian Prospective Diabetes Follow-Up Registry

Background: To analyze key indicators of metabolic control in adults with type 1 diabetes (T1D) using real-time or intermittent scanning continuous glucose monitoring (rtCGM/iscCGM) during real-life care, based on the German/Austrian/Swiss Prospective Diabetes Follow-up (DPV) registry. Methods: Cross-sectional analysis including 233 adults with T1D using CGM. We assessed CGM metrics by gender, age group (18 to <30 years vs. ≥30 years), insulin delivery method (multiple daily injections vs. continuous subcutaneous insulin infusion [CSII]) and sensor type (iscCGM vs. rtCGM), working days versus weekends, and daytime versus night-time using multivariable linear regression models (adjusted for demographic variables) or Wilcoxon signed-rank test. Results: Overall, 79/21% of T1D patients used iscCGM/rtCGM. Those aged ≥30 years spent more time in range (TIR [70-180 mg/dL] 54% vs. 49%) and hypoglycemic range <70 mg/dL (7% vs. 5%), less time in hyperglycemic range >180 mg/dL (38% vs. 46%) and had a lower glucose variability (coefficient of variation [CV] 36% vs. 37%) compared with adults aged <30 years. We found no significant differences between genders. Multivariable regression models revealed the highest Time In Range (TIR) and lowest time with sensor glucose >250 mg/dL, CV and daytime-night-time differences in those treated with CSII and rtCGM. Glucose profiles were slightly more favorable on working days. Conclusions: In our real-world data, rtCGM versus iscCGM was associated with a higher percentage of TIR and improved metabolic stability. Differences in ambulatory glucose profiles on working and weekend days may indicate lifestyle habits affecting glycemic stability. Real-life CGM results should be included in benchmarking reports in addition to hemoglobin A1c (HbA1c) and history of hypoglycemia.

Artificial Intelligence in Decision Support Systems for Type 1 Diabetes

Type 1 diabetes (T1D) is a chronic health condition resulting from pancreatic beta cell dysfunction and insulin depletion. While automated insulin delivery systems are now available, many people choose to manage insulin delivery manually through insulin pumps or through multiple daily injections. Frequent insulin titrations are needed to adequately manage glucose, however, provider adjustments are typically made every several months. Recent automated decision support systems incorporate artificial intelligence algorithms to deliver personalized recommendations regarding insulin doses and daily behaviors. This paper presents a comprehensive review of computational and artificial intelligence-based decision support systems to manage T1D. Articles were obtained from PubMed, IEEE Xplore, and ScienceDirect databases. No time period restrictions were imposed on the search. After removing off-topic articles and duplicates, 562 articles were left to review. Of those articles, we identified 61 articles for comprehensive review based on algorithm evaluation using real-world human data, in silico trials, or clinical studies. We grouped decision support systems into general categories of (1) those which recommend adjustments to insulin and (2) those which predict and help avoid hypoglycemia. We review the artificial intelligence methods used for each type of decision support system, and discuss the performance and potential applications of these systems.

The Use of a Smart Bolus Calculator Informed by Real-time Insulin Sensitivity Assessments Reduces Postprandial Hypoglycemia Following an Aerobic Exercise Session in Individuals With Type 1 Diabetes

OBJECTIVE

Insulin dosing in type 1 diabetes (T1D) is oftentimes complicated by fluctuating insulin requirements driven by metabolic and psychobehavioral factors impacting individuals' insulin sensitivity (IS). In this context, smart bolus calculators that automatically tailor prandial insulin dosing to the metabolic state of a person can improve glucose management in T1D.

RESEARCH DESIGN AND METHODS

Fifteen adults with T1D using continuous glucose monitors (CGMs) and insulin pumps completed two 24-h admissions in a hotel setting. During the admissions, participants engaged in an early afternoon 45-min aerobic exercise session, after which they received a standardized dinner meal. The dinner bolus was computed using a standard bolus calculator or smart bolus calculator informed by real-time IS estimates. Glucose control was assessed in the 4 h following dinner using CGMs and was compared between the two admissions.

RESULTS

The IS-informed bolus calculator allowed for a reduction in postprandial hypoglycemia as quantified by the low blood glucose index (2.02 vs. 3.31, P = 0.006) and percent time <70 mg/dL (8.48% vs. 15.18%, P = 0.049), without increasing hyperglycemia (high blood glucose index: 3.13 vs. 2.09, P = 0.075; percent time >180 mg/dL: 13.24% vs. 10.42%, P = 0.5; percent time >250 mg/dL: 2.08% vs. 1.19%, P = 0.317). In addition, the number of hypoglycemia rescue treatments was reduced from 12 to 7 with the use of the system.

CONCLUSIONS

The study shows that the proposed IS-informed bolus calculator is safe and feasible in adults with T1D, appropriately reducing postprandial hypoglycemia following an exercise-induced IS increase.